added new TextstyleParser implementation

* the old implementation was cumbersome to maintain and had some problems with nested formatting. see #630
2025-12-19 07:34:23 +01:00 · 2024-12-21 22:10:35 +01:00
parent 54c884c06e
commit f0a49e6aac
4 changed files with 166 additions and 114 deletions
--- a/2
+++ b/2
@@ -7,7 +7,7 @@ ARG GRAALVM_VERSION=21.0.0

 ARG BUILD_VERSION_ARG=unset

-FROM golang:1.22-bookworm AS buildcontainer
+FROM golang:1.23-bookworm AS buildcontainer

 ARG SIGNAL_CLI_VERSION
 ARG LIBSIGNAL_CLIENT_VERSION
--- a/src/client/client.go
+++ b/src/client/client.go
@@ -398,7 +398,8 @@ func (s *SignalClient) send(signalCliSendRequest ds.SignalCliSendRequest) (*Send

 	signalCliTextFormatStrings := []string{}
 	if signalCliSendRequest.TextMode != nil && *signalCliSendRequest.TextMode == "styled" {
-		signalCliSendRequest.Message, signalCliTextFormatStrings = utils.ParseMarkdownMessage(signalCliSendRequest.Message)
+		textstyleParser := utils.NewTextstyleParser(signalCliSendRequest.Message)
+		signalCliSendRequest.Message, signalCliTextFormatStrings = textstyleParser.Parse()
 	}

 	var groupId string = ""
--- a/src/utils/textstyleparser.go
+++ b/src/utils/textstyleparser.go
@@ -2,6 +2,8 @@ package utils

 import (
 	"strconv"
+	"unicode/utf16"
+	"unicode/utf8"
 )

 const (
@@ -18,119 +20,141 @@ const (
 	ItalicBegin            = 1
 	ItalicEnd              = 2
 	BoldBegin              = 3
-	BoldEnd1               = 4
-	BoldEnd2               = 5
 	MonoSpaceBegin         = 6
-	MonoSpaceEnd           = 7
 	StrikethroughBegin     = 8
-	StrikethroughEnd       = 9
-	SpoilerBegin1          = 10
-	SpoilerBegin           = 11
-	SpoilerEnd1            = 12
-	SpoilerEnd2            = 13
+	SpoilerBegin           = 9
 )

-func getUtf16CharacterCount(s string) int {
-	stringLength := len(s)
-	if stringLength == 1 {
-		return 1
+func getUtf16StringLength(s string) int {
+	runes := []rune(s) //turn string to slice
+
+	length := 0
+	for _, r := range runes {
+		length += utf16.RuneLen(r)
 	}
-	return stringLength / 2
+	return length
 }

-func getAdditionalCharacterCount(characterCount int) int {
-	additionalCharacterCount := characterCount - 1
-	if additionalCharacterCount > 0 {
-		return additionalCharacterCount
-	}
-	return 0
+type TokenState struct {
+	BeginPos int
+	Token    int
 }

-func ParseMarkdownMessage(message string) (string, []string) {
-	textFormat := Normal
-	textFormatBegin := 0
-	textFormatLength := 0
-	numOfControlChars := 0
-	state := None
-	signalCliFormatStrings := []string{}
-	fullString := ""
-	lastChar := ""
-	additionalCharacterCount := 0
+type Stack []TokenState

-	runes := []rune(message) //turn string to slice
+func (s *Stack) Push(v TokenState) {
+	*s = append(*s, v)
+}

-	for i, v := range runes { //iterate through rune
-		if v == '*' {
-			if state == ItalicBegin {
-				if lastChar == "*" {
-					state = BoldBegin
-					textFormat = Bold
-					textFormatBegin = i - numOfControlChars + additionalCharacterCount
-					textFormatLength = 0
-					additionalCharacterCount = 0
+func (s *Stack) Pop() TokenState {
+	ret := (*s)[len(*s)-1]
+	*s = (*s)[0 : len(*s)-1]
+
+	return ret
+}
+
+func (s *Stack) Peek() TokenState {
+	ret := (*s)[len(*s)-1]
+	return ret
+}
+
+func (s *Stack) Empty() bool {
+	if len(*s) == 0 {
+		return true
+	}
+	return false
+}
+
+const eof = -1
+
+type TextstyleParser struct {
+	input                  string
+	pos                    int
+	width                  int
+	tokens                 Stack
+	fullString             string
+	signalCliFormatStrings []string
+	//numOfControlTokens int
+}
+
+func NewTextstyleParser(input string) *TextstyleParser {
+	return &TextstyleParser{
+		input:                  input,
+		pos:                    0,
+		width:                  0,
+		tokens:                 make(Stack, 0),
+		fullString:             "",
+		signalCliFormatStrings: []string{},
+	}
+}
+
+func (l *TextstyleParser) next() (rune rune) {
+	if l.pos >= len(l.input) {
+		l.width = 0
+		return eof
+	}
+	//r := []rune(l.input[l.pos:])[0]
+	//l.width = utf16.RuneLen(r)
+	//l.pos += l.width
+	rune, l.width = utf8.DecodeRuneInString(l.input[l.pos:])
+	l.pos += l.width
+	return rune
+}
+
+// backup steps back one rune.
+// Can be called only once per call of next.
+func (l *TextstyleParser) backup() {
+	l.pos -= l.width
+}
+
+// peek returns but does not consume
+// the next rune in the input.
+func (l *TextstyleParser) peek() rune {
+	rune := l.next()
+	l.backup()
+	return rune
+}
+
+func (l *TextstyleParser) handleToken(tokenType int, signalCliStylingType string) {
+	if l.tokens.Empty() {
+		l.tokens.Push(TokenState{BeginPos: getUtf16StringLength(l.fullString), Token: tokenType})
 	} else {
-					state = ItalicEnd
-				}
-			} else if state == None {
-				state = ItalicBegin
-				textFormat = Italic
-				textFormatBegin = i - numOfControlChars + additionalCharacterCount
-				textFormatLength = 0
-			} else if state == BoldBegin {
-				state = BoldEnd1
-			} else if state == BoldEnd1 {
-				state = BoldEnd2
-			}
-			numOfControlChars += 1
-		} else if v == '|' {
-			if state == None {
-				state = SpoilerBegin1
-			} else if state == SpoilerBegin1 && lastChar == "|" {
-				state = SpoilerBegin
-				textFormat = Spoiler
-				textFormatBegin = i - numOfControlChars + additionalCharacterCount
-				textFormatLength = 0
-			} else if state == SpoilerBegin {
-				state = SpoilerEnd1
-			} else if state == SpoilerEnd1 && lastChar == "|" {
-				state = SpoilerEnd2
-			}
-			numOfControlChars += 1
-		} else if v == '`' {
-			if state == None {
-				state = MonoSpaceBegin
-				textFormat = Monospace
-				textFormatBegin = i - numOfControlChars + additionalCharacterCount
-				textFormatLength = 0
-			} else if state == MonoSpaceBegin {
-				state = MonoSpaceEnd
-			}
-			numOfControlChars += 1
-		} else if v == '~' {
-			if state == None {
-				state = StrikethroughBegin
-				textFormat = Strikethrough
-				textFormatBegin = i - numOfControlChars + additionalCharacterCount
-				textFormatLength = 0
-			} else if state == StrikethroughBegin {
-				state = StrikethroughEnd
-			}
-			numOfControlChars += 1
+		if l.tokens.Peek().Token == tokenType {
+			tokenBeginState := l.tokens.Pop()
+			l.signalCliFormatStrings = append(l.signalCliFormatStrings, strconv.Itoa(tokenBeginState.BeginPos)+":"+strconv.Itoa(getUtf16StringLength(l.fullString)-tokenBeginState.BeginPos)+":"+signalCliStylingType)
 		} else {
-			textFormatLength += 1
-			fullString += string(v)
-			additionalCharacterCount += getAdditionalCharacterCount(getUtf16CharacterCount(string(v)))
-		}
-		lastChar = string(v)
-
-		if state == ItalicEnd || state == BoldEnd2 || state == MonoSpaceEnd || state == StrikethroughEnd || state == SpoilerEnd2 {
-			signalCliFormatStrings = append(signalCliFormatStrings, strconv.Itoa(textFormatBegin)+":"+strconv.Itoa(textFormatLength+additionalCharacterCount)+":"+textFormat)
-			state = None
-			textFormatBegin = 0
-			textFormatLength = 0
-			textFormat = Normal
+			l.tokens.Push(TokenState{BeginPos: getUtf16StringLength(l.fullString), Token: tokenType})
 		}
 	}
-
-	return fullString, signalCliFormatStrings
+}
+
+func (l *TextstyleParser) Parse() (string, []string) {
+	for {
+		c := l.next()
+		if c == eof {
+			break
+		}
+
+		nextRune := l.peek()
+
+		if c == '*' {
+			if nextRune == '*' { //Bold
+				l.next()
+				l.handleToken(BoldBegin, Bold)
+			} else { //Italic
+				l.handleToken(ItalicBegin, Italic)
+			}
+		} else if (c == '|') && (nextRune == '|') {
+			l.next()
+			l.handleToken(SpoilerBegin, Spoiler)
+		} else if c == '~' {
+			l.handleToken(StrikethroughBegin, Strikethrough)
+		} else if c == '`' {
+			l.handleToken(MonoSpaceBegin, Monospace)
+		} else {
+			l.fullString += string(c)
+		}
+	}
+
+	return l.fullString, l.signalCliFormatStrings
 }
--- a/src/utils/textstyleparser_test.go
+++ b/src/utils/textstyleparser_test.go
@@ -15,80 +15,107 @@ func expectFormatStringsEqual(t *testing.T, formatStrings1 []string, formatStrin
 	}
 }

-func TestSimpleMessage1(t *testing.T) {
-	message, signalCliFormatStrings := ParseMarkdownMessage("*italic*")
+func TestSimpleItalicMessage(t *testing.T) {
+	textstyleParser := NewTextstyleParser("*italic*")
+	message, signalCliFormatStrings := textstyleParser.Parse()
 	expectMessageEqual(t, message, "italic")
 	expectFormatStringsEqual(t, signalCliFormatStrings, []string{"0:6:ITALIC"})
 }

+func TestSimpleBoldMessage(t *testing.T) {
+	textstyleParser := NewTextstyleParser("**bold**")
+	message, signalCliFormatStrings := textstyleParser.Parse()
+	expectMessageEqual(t, message, "bold")
+	expectFormatStringsEqual(t, signalCliFormatStrings, []string{"0:4:BOLD"})
+}
+
 func TestSimpleMessage(t *testing.T) {
-	message, signalCliFormatStrings := ParseMarkdownMessage("*This is a italic message*")
+	textstyleParser := NewTextstyleParser("*This is a italic message*")
+	message, signalCliFormatStrings := textstyleParser.Parse()
 	expectMessageEqual(t, message, "This is a italic message")
 	expectFormatStringsEqual(t, signalCliFormatStrings, []string{"0:24:ITALIC"})
 }

 func TestBoldAndItalicMessage(t *testing.T) {
-	message, signalCliFormatStrings := ParseMarkdownMessage("This is a **bold** and *italic* message")
+	textstyleParser := NewTextstyleParser("This is a **bold** and *italic* message")
+	message, signalCliFormatStrings := textstyleParser.Parse()
 	expectMessageEqual(t, message, "This is a bold and italic message")
 	expectFormatStringsEqual(t, signalCliFormatStrings, []string{"10:4:BOLD", "19:6:ITALIC"})
 }

 func TestTwoBoldFormattedStrings(t *testing.T) {
-	message, signalCliFormatStrings := ParseMarkdownMessage("This is a **bold** and another **bold** message")
+	textstyleParser := NewTextstyleParser("This is a **bold** and another **bold** message")
+	message, signalCliFormatStrings := textstyleParser.Parse()
 	expectMessageEqual(t, message, "This is a bold and another bold message")
 	expectFormatStringsEqual(t, signalCliFormatStrings, []string{"10:4:BOLD", "27:4:BOLD"})
 }

 func TestStrikethrough(t *testing.T) {
-	message, signalCliFormatStrings := ParseMarkdownMessage("This is a ~strikethrough~ and a **bold** message")
+	textstyleParser := NewTextstyleParser("This is a ~strikethrough~ and a **bold** message")
+	message, signalCliFormatStrings := textstyleParser.Parse()
 	expectMessageEqual(t, message, "This is a strikethrough and a bold message")
 	expectFormatStringsEqual(t, signalCliFormatStrings, []string{"10:13:STRIKETHROUGH", "30:4:BOLD"})
 }

 func TestMonospace(t *testing.T) {
-	message, signalCliFormatStrings := ParseMarkdownMessage("This is a `monospace` and a **bold** message")
+	textstyleParser := NewTextstyleParser("This is a `monospace` and a **bold** message")
+	message, signalCliFormatStrings := textstyleParser.Parse()
 	expectMessageEqual(t, message, "This is a monospace and a bold message")
 	expectFormatStringsEqual(t, signalCliFormatStrings, []string{"10:9:MONOSPACE", "26:4:BOLD"})
 }

 func TestMulticharacterEmoji(t *testing.T) {
-	message, signalCliFormatStrings := ParseMarkdownMessage("👋abcdefg")
+	textstyleParser := NewTextstyleParser("👋abcdefg")
+	message, signalCliFormatStrings := textstyleParser.Parse()
 	expectMessageEqual(t, message, "👋abcdefg")
 	expectFormatStringsEqual(t, signalCliFormatStrings, []string{})
 }

 func TestMulticharacterEmojiWithBoldText(t *testing.T) {
-	message, signalCliFormatStrings := ParseMarkdownMessage("👋**abcdefg**")
+	textstyleParser := NewTextstyleParser("👋**abcdefg**")
+	message, signalCliFormatStrings := textstyleParser.Parse()
 	expectMessageEqual(t, message, "👋abcdefg")
 	expectFormatStringsEqual(t, signalCliFormatStrings, []string{"2:7:BOLD"})
 }

 func TestMultipleMulticharacterEmoji(t *testing.T) {
-	message, signalCliFormatStrings := ParseMarkdownMessage("👋🏾abcdefg")
+	textstyleParser := NewTextstyleParser("👋🏾abcdefg")
+	message, signalCliFormatStrings := textstyleParser.Parse()
 	expectMessageEqual(t, message, "👋🏾abcdefg")
 	expectFormatStringsEqual(t, signalCliFormatStrings, []string{})
 }

 func TestMultipleMulticharacterEmojiWithBoldText(t *testing.T) {
-	message, signalCliFormatStrings := ParseMarkdownMessage("👋🏾**abcdefg**")
+	textstyleParser := NewTextstyleParser("👋🏾**abcdefg**")
+	message, signalCliFormatStrings := textstyleParser.Parse()
 	expectMessageEqual(t, message, "👋🏾abcdefg")
 	expectFormatStringsEqual(t, signalCliFormatStrings, []string{"4:7:BOLD"})
 }

 func TestMulticharacterEmojiWithBoldText2(t *testing.T) {
-	message, signalCliFormatStrings := ParseMarkdownMessage("Test 👦🏿 via **signal** API")
+	textstyleParser := NewTextstyleParser("Test 👦🏿 via **signal** API")
+	message, signalCliFormatStrings := textstyleParser.Parse()
 	expectMessageEqual(t, message, "Test 👦🏿 via signal API")
 	expectFormatStringsEqual(t, signalCliFormatStrings, []string{"14:6:BOLD"})
 }

 func TestSpoiler(t *testing.T) {
-	message, signalCliFormatStrings := ParseMarkdownMessage("||this is a spoiler||")
+	textstyleParser := NewTextstyleParser("||this is a spoiler||")
+	message, signalCliFormatStrings := textstyleParser.Parse()
 	expectMessageEqual(t, message, "this is a spoiler")
 	expectFormatStringsEqual(t, signalCliFormatStrings, []string{"0:17:SPOILER"})
 }

 func TestSpoiler1(t *testing.T) {
-	message, signalCliFormatStrings := ParseMarkdownMessage("||this is a spoiler|| and another ||spoiler||")
+	textstyleParser := NewTextstyleParser("||this is a spoiler|| and another ||spoiler||")
+	message, signalCliFormatStrings := textstyleParser.Parse()
 	expectMessageEqual(t, message, "this is a spoiler and another spoiler")
 	expectFormatStringsEqual(t, signalCliFormatStrings, []string{"0:17:SPOILER", "30:7:SPOILER"})
 }
+
+func TestBoldTextInsideSpoiler(t *testing.T) {
+	textstyleParser := NewTextstyleParser("||**this is a bold text inside a spoiler**||")
+	message, signalCliFormatStrings := textstyleParser.Parse()
+	expectMessageEqual(t, message, "this is a bold text inside a spoiler")
+	expectFormatStringsEqual(t, signalCliFormatStrings, []string{"0:36:BOLD", "0:36:SPOILER"})
+}